Case–control studies

2.2.1 Cancer of the lung (Table 2.2)

In 1989 (Monograph 47), nine case–control studies of lung cancer and two multisite case–control studies, which included lung cancer, were evaluated. These studies are summarized in Table 24 of Monograph 47 (IARC, 1989).

(a) Europe

Jahn et al. (1999) carried out a pooled analysis of the two case–control studies on lung cancer conducted in Germany: the Bremen Institute for Prevention Research and Social Medicine (BIPS) study in the Bremen and Frankfurt/Main areas, during 1988–1993, and the GSF-National Research Center for Environment and Health (GSF) study in Nordrhein-Westfalen, Rheinland-Pfalz and Bayern, Saarland, Thuringen, and Sachsen, during 1990–

1996. The results from the BIPS study had been reported earlier by Jöckel et al. (1992, 1998) for both sexes combined. The Jahn et al. (1999) analysis was restricted to women, and included 686 cases aged 75 or less at diagnosis, of German nationality, residing in the study regions. All cases were confirmed by histology or cytology. Population controls, 712 individuals, were randomly selected from population registries or by random digit dialling, and were individually (BIPS study) or frequency- (GSF study) matched to cases by age, and region. A standardized questionnaire, with full occupational history and supplementary job-specific modules, was administered during face-to-face interviews. The response rate was 73% among cases, and 45% in controls. An odds ratio (OR) of 3.00 (95% CI: 0.73–12.33) was found after adjustment for smoking and asbestos exposure (age and region of residence were strata-defining variables in the conditional logistic regression models) for the occupation of ‘ever’ painter. [A major strength of this study was exposure definitions, based on complete and accurate occupational histories, and expert-based quantitative exposure assessment for a series of carcinogens. However, the low response rate among controls might have led to selection bias].

Brüske-Hohlfeld et al. (2000) also carried out a pooled analysis of the two case-control studies described above (the BIPS and GSF studies). The results from the BIPS study had been reported earlier by Jöckel et al. (1992, 1998) for both sexes combined. This analysis was restricted to men, and included 3498 cases aged 76 or less at diagnosis who lived in Germany for at least 25 years and resided in the study regions. All cases were confirmed by histology or cytology. Population controls, 3541 individuals, were randomly selected from population registries or by random digit dialling, and were individually (BIPS study) or

188 IARC MONOGRAPHS VOLUME 98

frequency- (GSF study) matched to cases by age and region. A standardized questionnaire, with full occupational history and supplementary job-specific modules, was administered during face-to-face interviews. The response rate was 77% among cases, 41% in controls.

An OR of 1.42 (95% CI: 1.05–1.92) was found after adjustment for smoking and asbestos exposure (age class and region of residence were strata-defining variables in the conditional logistic regression modelling) for the occupation of “ever painter/lacquerer.” [A major strength of this study was the exposure definition, based on complete and accurate occupational histories, aiming to expert-based quantitative exposure assessment to a series of carcinogens. However, the low response rate among controls might have led to selection bias].

Pohlabeln et al. (2000) conducted a case–control study among non-smokers in 12 European study centres in Germany, Italy, Portugal, Sweden, United Kingdom, France and Spain to evaluate the role of occupational risk factors among non-smokers. Non-smoking cases and controls were defined as subjects who smoked fewer than 400 cigarettes during their lifetime. Lifetime occupational histories in face-to-face interviews were obtained from 650 non-smoking cases (509 females, 141 males) and 1542 non-smoking controls (1011 females, 141 males). Community-based controls were selected in six centres, hospital-based controls in five centres, and both community and hospital-based controls in one centre. Hospital controls were selected from diseases not related to tobacco smoking.

Painting was among the three occupations where an excess risk was identified in males (OR, 1.84; 95% CI: 0.59–5.74; based on six cases). Numbers of females involved were too small to produce reliable estimates of effect. [This is the only case–control study of non-smokers, sufficiently large to study occupational exposures.]

Bouchardy et al. (2002) identified 58 134 incident cancer cases in men from five cantonal Swiss Cancer Registries (Basel, Geneva, St Gall, Vaud, and Zurich), 1980–9193.

The overall proportion with histological or cytological confirmation of diagnosis was 95.1%. The study was restricted to cases aged 25 years or more at registration (and less than 65 year in St Gall and Vaud). The longest, current or most recent occupation at registration was recorded (the main or most accurately specified occupation was used in the Zurich Registry). Subjects with unknown occupation were not reported separately. The association between different cancer sites and work in a pre-defined set of industries and occupation was studied by estimating ORs adjusted for age, registry, civil status, period of diagnosis, nationality, urban/rural residence, and socioeconomic status. For each neoplasm, registrants for the other cancer sites were used as reference. Overall, 9106 lung cancer cases were registered, 273 of those were plasterers and painters in the construction industry. A total of 49 028 non-lung cancer cases were registered, 867 of whom were painters. The OR for cancer of the lung among painters was 1.1 (95% CI: 1.0–1.3), adjusted for all variables. The OR when adjusted for all variables except socioeconomic status was 1.4 (95% CI: 1.2–1.6).

[ORs could not be adjusted for smoking.]

Richiardi et al. (2004) carried out a population-based case-control study in two industrialized areas of Northern Italy (city of Turin and Eastern Veneto) in 1990–1992, including 1132 lung cancer cases and 1553 controls less than 75 years of age. Histologically

OCCUPATIONAL EXPOSURE AS A PAINTER 189 or cytologically confirmed cases were identified through weekly monitoring of all hospitals in the study areas. Population controls were frequency-matched with cases by sex, study area and 5-year age groups. Response proportions for Turin and Eastern Veneto were, respectively, 86% and 72% among cases, and, respectively, 85% and 74% among controls.

A face-to-face interview was used to collect information on each subject’s occupation lasting >6 months, record the job title and industry and the time period of employment.

Occupational histories were coded according to international classifications and evaluated for employment in occupations known (list A) or suspected (list B) to determine any exposure to lung carcinogens. This was done using a previously suggested translation of lists A and B into combinations of codes for job titles and industries (Ahrens & Merletti, 1998). Analyses on specific list A occupations, including painters, was limited to men (956 cases and 1253 controls). Compared to men who were never employed in occupations in lists A or B, painters had an OR for cancer of the lung of 2.0 (95% CI: 1.4–3.3) after adjusting for matching variables, smoking and number of job periods, and of 1.7 (95% CI:

1.1–3.0) if additionally adjusting for educational level.

Baccarelli et al. (2005) conducted a study in the Leningrad province (the Russian Federation), during 1993–1998 on lung cancer cases diagnosed at autopsy: 540 cases (474 men, 66 women) diagnosed at postmortem examination at the St Petersburg central pathology laboratory, serving 88 state hospitals in the study area, were included along with 582 (453 men, 129 women) individuals with diagnoses of non-cancer, non-tobacco related conditions, frequency-matched by sex, age, area, year of death. Postmortem examinations were conducted in about 95% of decedents in the state hospitals involved. Full occupational records were retrieved for all cases and all controls. Information on smoking was abstracted from medical records at local health centres; however the proportion of success for data abstraction and data quality are not stated. An OR of 0.6 (95% CI: 0.3–1.4) was found for occupation as ‘ever painter’, 0.5 (95% CI: 0.2–1.5) for < 10 years of employment as a painter, and 0.8 (95% CI: 0.2–3.0) for ≥ 10 years of employment as a painter, adjusted for age, sex, and smoking.

Zeka et al. (2006) conducted a multicentre case–control study of lung cancer in several European countries between 1998–2002. A total of 223 ‘never’ smoking cases (48 men, 175 women) and 1039 non-smoking controls (534 men, 505 women) were included in the analysis. In-person interviews were conducted to obtain lifetime occupational histories for jobs held >1 year. Occupation as a painter was associated with a non-significant increased risk of lung cancer among women (OR, 1.8; 95% CI: 0.53–6.0, based on six cases and six controls), adjusted for age and study centre.

190 IARC MONOGRAPHS VOLUME 98 Table 2.2. Case-control studies of lung cancer among persons with occupation as a painter

Reference, Brüske-Hohlfeld et al.

(2000) in Germany for at least 25 years, resident in the study region 100% confirmed by histology or cytology.

Response rate 63% BIPS, 77%

GSF [73%

overall]

712 female and 3541 male registries or by random-digit according to the classification of

Low response rate among controls with potential for selection bias;

frequency matched cases and controls of the GSF-study were post-hoc stratified according to the matching variables age, region; *fixed effects model used to calculate a weighted average;

these studies have substantial overlap with Kreuzer et al.

(2001) that presented results for painters in lifetime non-smoking men [2.31 (0.57–9.47)]

and women (OR=1.2), respectively. BIPS study overlaps with Jöckel et al. (1998)